California Congressional Primary Analysis Part I

The first step to determining the results in the fall elections in California is understanding the relationship between the primary electorate and the general election electorate. Many people understand that the primary electorate was more Republican than the general election will likely be. The primary was 43%D/37%R/20%I. The general election electorate in California closely mirrors registration, according to exit polls. Registration is 43%D/30%R/27%I.

Things become problematic when people make the assumption that since the November election will be more Democratic/less Republican and thus Democrats will compete in a lot of districts. The first assumption is probably true in most, but not all, cases, and the second assumption applies the first assumption equally everywhere.

Unfortunately, the data needed to study those assumptions isn’t readily available. The Secretary of State has provided party participation numbers by county for the primary in the past, but has yet to do so this year. They don’t break this data down by congressional, assembly, or senate districts. They provide no party information for the general election.

Fortunately, many of the California counties do provide the information broken down by party, although some use a format that doesn’t. By going county by county, I do have party participation numbers for those counties. For the others I used total votes in the Presidential primary. This method overestimates Democrats, because independents are allowed to participate in their Presidential primary. Past primaries have shown independents to make up less than 5% of the Democratic electorate, however, and this Presidential primary wasn’t likely to draw many crossover voters. So the increase may have been negligible. Secondarily, the counties usually doesn’t count under votes, ballots where the voter left the Presidential box empty or didn’t fill it in properly. There are more of these than you’d think, especially considering that Democrats unhappy with Barack Obama didn’t have a lot of choice to lodge a protest vote.

Before we get into this year’s numbers, let’s look at the 2010 numbers. Since the counties don’t break turnout down by congressional district, I’ve had to apply the county numbers to each of the districts within that county. That’s not exact, because there’s no reason to think all the districts have the same turnout characteristics. Still, I was able to get an approximation of how the primary vote/participation and general election vote/participation went for the more competitive districts.

CA-3, now largely in CA-7, is in Sacramento county. At the time of the June 2010 primary, Democrats had a 13.3% registration advantage in the county. Yet they participated only 6.2% higher. If the general election electorate moved to registration you’d expect a similar move in the primary to general vote. Dan Lungren won the primary by 18.6%, but only won the general election by 6.9%. So the general moved further in the Democratic direction than you’d expect.

All in all, 6 of these 11 competitive districts didn’t move as Democratic as you’d expect, but 5 moved more Democratic than expected in 2010. In most districts there wasn’t a consistent election to election pattern. CA-3 in 2006 had 2.7% higher Republican participation than registration in the primary, but the general actually went the other way. Lungren won the primary by 13.9% and won the general by 19.6%. For most districts I’ll assume that the results will move similarly to registration.

The one exception to this is the Central Valley. In the two Central Valley districts in 2010, Republican participation was dramatically higher in the primary than registration, but the move wasn’t nearly as big in the general. In fact, CA-20 moved heavily the other way. This confirms that the Central Valley has a lot of registered Democrats that don’t vote. So I wouldn’t expect the new Central Valley districts to vote closer to the registration in the general election.